Design Considerations

Modeling the Effects of

Soil-Structure Interaction
on a Tall Building
Bearing on a Mat
Foundation
Examining soil-structure
interaction can result in
more efficient structural
designs.
LEWIS EDGERS, MASOUD SANAYEI &
JOSEPH L. ALONGE

he interaction of a building, its foundation and the underlying soils may have
important effects on the behavior of
each of these components as well as on the
overall system behavior. For example, the relative stiffness of a building structure, its mat
foundation and the soils that support the
foundation will influence the stresses and displacements of both the structure and soil.
Soil-structure interaction (SSI) effects are
sometimes neglected by the use of a structural model supported on a fixed base. Other

simple models assume an ideally flexible or

infinitely rigid foundation on an elastic subsoil.
An investigation of the effects of SSI on the
stresses and displacements in the structure
and the soil of a model fifty-story steel frame
structure with a concrete mat foundation
bearing on a deformable soil was undertaken
as a means of best understanding how to perform and apply SSI analyses. The study
included investigating the effects of the stiffnesses of the building, its mat foundation and
an elastic subsoil on the stresses, internal
forces and displacements of the building,
foundation and subsoils. The soil-structure
model also considered the effects of foundation embedment.

Development of
Computer Models
The study of SSI was conducted with a building frame model developed within the context
of research on the effects of increased wind

CIVIL ENGINEERING PRACTICE

FALL/WINTER 2005

51

FIGURE 1. ANSYS model of the fifty-story

building pin-supported on a rigid base.
loads due to global climate change on tall
1
buildings in Boston. The building model is
representative of many taller buildings found
in large cities.

52

CIVIL ENGINEERING PRACTICE

FALL/WINTER 2005

For the initial analyses, an ANSYS model of

the building frame was pin-connected to a
rigid base (see Figure 1). The two-dimensional building model 182.9 meters (600 feet)
high by 30.5 meters (100 feet) wide by 0.3
meters (1 foot) consists of 255 nodes and
550 elements. The framing system is based on
a steel member with core bracing similar to
that of a Pratt brace. The floor height is 3.66
meters (12 feet) and is typical of many tall
2
buildings. It uses the Beam3 as frame elements and Link1 as truss elements. The framing material is assumed to be steel and the
flooring system is assumed to be metal decking with a concrete slab 10.2 centimeters (4
inches) thick. The wind loads were applied as
point loads along the left side of the structure,
and the vertical dead and live loads were
applied together as uniform loads to the horizontal members. Gravity loads actuate the
self-weight of the structure. The connection
between the building and mat foundation was
modeled by a pin connection at the base of the
columns. Figure 2 shows a close-up of the
building frame, lateral load carrying system
and the pin connections to a rigid support.
The structure is fixed at the base in order to
examine its response (since the structure is
ordinarily analyzed by ignoring the SSI).
Figure 3 shows the finite-element model of
the building frame supported by a mat foundation bearing on a stiff linear elastic soil. This
model was developed in order to simulate
realistic foundation design conditions. The
soil mass was represented by a mesh 213
meters (700 feet) wide by 152 meters (500 feet)
deep, consisting of forty-two quadrilateral
plane strain elements. A set of elements and
nodes are added to the top of the soil mass to
represent the concrete mat. The coincident
nodes at the soil-concrete interface were
merged so that there is full continuity with no
slippage at this interface along the base of the
foundation.
A two-dimensional, rather than threedimensional, structural model was developed
in order to increase computational efficiency.
In order to represent a three-dimensional
structure in two dimensions, the element stiffness and loadings were distributed over a unit
two-dimensional slice of frame bents spaced

FIGURE 2. The lower floors of a fifty-story building pin-supported on a rigid base.

at 7.6 meters (25 feet). The bending rigidity, EI,
and the axial rigidity, EA, of the buildings
beams and columns, located at the 7.6-meter
(25-foot) bay spacing, were scaled by reducing
the common elastic modulus parameter, E, by
one-twenty-fifth. Thus, it was unnecessary to
adjust the moment of inertia, I, or cross-sectional area, A, of each member. The material
density for the discrete members was scaled
by one-twenty-fifth so that full self-weight
was evenly applied over each bay. No scaling
was applied to the properties of the concrete
mat foundation and the subsoil because they
were continuous. Table 1 summarizes the
material properties used for these analyses. A
two-dimensional model would not be applicable for situations without symmetry in the
third dimension. For these situations, full
three-dimensional analyses can be performed
with finite-element packages (such as ANSYS
and PLAXIS 3D Foundation).
The dead, live and wind loads used in this
analysis are based on the Massachusetts State
3
Building Code. Seismic loads are not considered. Dead load includes the weight of the
framing, flooring, ceiling and mechanical, and
partition load (assumed to be office space).
The live load was assumed to be an average of
the typical loads found in offices, office lob-

bies and corridors. The wind loads are based

on Zone 3 for the Boston metropolitan area,
and exposure B for towns and cities. All
applied dead and live loads were transformed
to represent the forces acting on a unit slice of
the building model. To do so requires a onetwenty-fifth scaling of the weight of all steel
elements spaced at 7.6 meters (25 feet) but no
adjustment of distributed loads. The wind
loads were concentrated at each of the buildings floor levels. Tables 2 and 3 summarize
the loads applied to the unit slice model.
Figure 4 shows the details of the finite-element model, modified to describe an embedded foundation. The computer model of the
excavated areas includes both soil elements
and elements representing foundation walls,
mat foundation and two stories of an underground garage structure. The linear elastic
model requires full continuity between the soil
and the concrete walls and mat. The embedded
model also includes two levels of columns and
concrete beams to represent the underground
part of the structure. The entire mesh includes
7,083 nodes and 6,948 elements. For the purpose of these linear-elastic analyses, excavation
support issues have been neglected.
The ANSYS analyses for staged excavation
and construction were performed as follows:

CIVIL ENGINEERING PRACTICE

FALL/WINTER 2005

53